SoundPalette Extractor

The following are step-by-step instructions to install and run the SoundPalette extractor in a context where you can copy your archive of sounds into a source folder

(if you want to run the extractor on a folder on an external drive, check out this guide)

first, lets install docker!
- https://docs.docker.com/docker-for-mac/install/
now open the terminal app
- press command+spacebar to open spotlight, type terminal and press enter
navigate to your source-code folder
- if you don't have one, open the terminal app, and type the following commands to make a folder inside of your home folder:
  - ```
  cd ~/
  mkdir src
  cd src
```
let's verify that git is installed
- type git version and press enter

great, now clone the project repo with the following git command

git clone git@github.com:philipkobernik/sound-palette-extractor.git
cd sound-palette-extractor

now its time to start the docker image

docker run --rm -p 8888:8888 --mount type=bind,source=$(pwd),target=/notebooks mtgupf/mir-toolbox

if successful, we should see something like this:

[I 01:49:29.963 NotebookApp] Writing notebook server cookie secret to /root/.local/share/jupyter/runtime/notebook_cookie_secret
[I 01:49:31.404 NotebookApp] Serving notebooks from local directory: /notebooks
[I 01:49:31.405 NotebookApp] The Jupyter Notebook is running at:
[I 01:49:31.405 NotebookApp] http://(6cd0a12dffe2 or 127.0.0.1):8888/?token=...

open the notebook in your web browser by navigating to http://localhost:8888?token=mir
click on the file called palette-extractor.ipynb to open the notebook
the extractor looks for your audio files inside of the folder called corpus
- please move all of your archive audio files here (let me know if this is not possible)
- nested folder structures are supported
click Cell > Run All
The extractor may take some hours to complete, depending on the size of your archive
- on my 2016 macbook pro, the extractor runs about 2 files/minute
- for my archive of 941 files, it will run for about 8 hours
The notebook is not very efficient with memory, and memory usage will blow up like a balloon when processing large archives.
- for this reason, I recommend allocating 7+ GB of memory to the docker container
- if the notebook runs out of memory, the kernel will simply "die" and stop
  - if this happens, simply re-start the analysis process by clicking Cell > Run All
  - the algorithm will skip files that it has already processed